Correlations between ultrafast power Doppler perfusion imaging variables and clinical disease activity in rheumatoid arthritis: potential applications for diagnosing and treating patients in deep clinical remission

Article information

Ultrasonography. 2024;43(6):478-489
Publication date (electronic) : 2024 September 2
doi : https://doi.org/10.14366/usg.24095
1Graduate Institute of Biomedical Electronics and Bioinformatics, National Taiwan University, Taipei, Taiwan
2Division of Allergy, Immunology and Rheumatology, Department of Internal Medicine, Taichung Veterans General Hospital, Taichung, Taiwan
3Department of Post-Baccalaureate Medicine, College of Medicine, National Chung Hsing University, Taichung, Taiwan
4Department of Electrical Engineering, National Taiwan University, Taipei, Taiwan
Correspondence to: Pai-Chi Li, PhD, Department of Electrical Engineering, National Taiwan University, No. 1, Sec. 4, Roosevelt Rd., Taipei 10617, Taiwan (R.O.C.) Tel. +886-2-33663551 Fax. +886-2-23627651 E-mail: paichi@ntu.edu.tw
Received 2024 May 24; Revised 2024 August 26; Accepted 2024 September 2.

Abstract

Purpose

This study aimed to evaluate the ability of ultrafast power Doppler (PD) to assess disease activity in rheumatoid arthritis (RA) by examining the correlations between variables from ultrafast PD perfusion imaging and clinical measures of disease activity.

Methods

Thirty-three RA patients underwent clinical assessments of disease activity and ultrasound scans of bilateral wrists using both ultrafast and conventional PD systems. A spatial singular value decomposition filter was applied to the ultrafast PD imaging. Singular vectors representing perfusion and fast flows were selected to produce perfusion images. All images were quantitatively analyzed with computer assistance and scored semiquantitatively (0-3) by a physician for synovial vascularity. The Pearson correlation coefficients between image variables and clinical indices were calculated.

Results

The correlation coefficients ranged from weakly to moderately positive between ultrafast PD variables and clinical indices (r=0.221-0.374, all P<0.05). The strongest correlations were observed for synovial PD brightness with the 28-joint Disease Activity Score based on C-Reactive Protein (DAS28-CRP) and the Simplified Disease Activity Index (SDAI). In patients within the deep clinical remission (dCR) subgroup, synovial PD brightness showed stronger correlations with DAS28-CRP, the Clinical Disease Activity Index, and SDAI (r=0.578-0.641, all P<0.001). The correlation coefficients between conventional PD variables and clinical indices were similar to those observed with ultrafast PD variables.

Conclusion

Ultrafast PD imaging effectively extracts capillary blood signals and generates perfusion images. In the RA population, ultrafast PD variables exhibit weak-to-moderate correlations with clinical indices, with these correlations being notably stronger in dCR patients.

Graphic Abstract

Introduction

Rheumatoid arthritis (RA) is an autoimmune disorder characterized by inflammation of the synovium. The inflamed synovium undergoes hypertrophy, exhibiting lymphocyte infiltration and angiogenesis [1]. Musculoskeletal ultrasonography has become widely used to assess RA, and its sensitivity in detecting synovitis is comparable to that of magnetic resonance imaging [2]. Grayscale (GS) imaging reveals structural abnormalities in joints, while power Doppler (PD) imaging detects blood flow, indicating the extent of tissue inflammation [3]. In the early stages of synovitis, an increase in synovial vascularity appears before synovial hypertrophy. However, the capillary blood flow velocity, estimated at 0.3 mm/s, may not be detectable with conventional ultrasound systems due to the low sensitivity of the Doppler function [4]. This limitation can lead to an underestimation of RA disease activity.

Ultrafast Doppler is an innovative Doppler technique that emits plane waves and receives returning echoes at a high pulse-repetition frequency (PRF). This technique utilizes a specially designed emission program with compounding, which extends the duration of flow observations. This enhancement improves velocity resolution and the ability to detect very slow flows compared to conventional Doppler, which uses focused transmit beams [5]. The Doppler function can be further improved by applying a singular value decomposition (SVD) filter [6,7]. Ultrafast PD equipped with an SVD filter can detect significantly slower flows than ultrafast PD using a standard wall filter, offering potential benefits in imaging synovial microvascularity in RA [7]. The authors’ recent study demonstrated that ultrafast PD with SVD filtering exhibited markedly higher sensitivity (0.985 vs. 0.424) and accuracy (0.813 vs. 0.604) in diagnosing RA than conventional PD [8].

Conventional PD, interpreted using the Outcome Measures in Rheumatology (OMERACT) ultrasound scoring system with a semiquantitative score ranging from 0 to 3, is commonly employed to assess RA disease activity [9]. However, the OMERACT PD score for a single joint shows only a weak correlation with clinical indices [10]. Specifically, conventional PD struggles to detect slow blood flows in cases of low-grade inflammation, resulting in an underestimation of synovitis severity [11]. Since ultrafast PD is recognized for its superior sensitivity in detecting synovial microvascularity, this study explored its effectiveness in evaluating RA disease activity. This study aimed to establish correlations between ultrafast PD perfusion imaging findings and RA clinical disease activity, and to assess the clinical utility of ultrafast PD in this context.

Materials and Methods

Compliance with Ethical Standards

This prospective and observational study received approval from Taichung Veterans General Hospital Institutional Review Board (protocol number: CF19053B). Informed consent was obtained from each participant.

Patient Selection

The inclusion criteria required participants to be at least 20 years old and to have a diagnosis of RA, as defined by the 2010 American College of Rheumatology/European League Against Rheumatism (ACR/EULAR) classification criteria for RA [12]. The exclusion criteria included the presence of concomitant crystal-induced or septic arthritis at the wrist, as well as severe wrist deformity that could interfere with the ultrasound assessment. Between November 2020 and October 2021, thirty-five RA patients were screened, and 33 eligible patients were subsequently enrolled. These patients had an average age of 49.2±12.4 years (range, 21 to 71 years) and were predominantly female (84.8%).

Thirty-two patients were on disease-modifying antirheumatic drugs such as methotrexate, hydroxychloroquine, sulfasalazine, leflunomide, or a combination thereof, while five patients were treated with biologics (four with anti-tumor necrosis factor alpha and one with tofacitinib monotherapy) at the time of enrollment. They continued their regular medications throughout the study. These patients underwent clinical and ultrasound assessments, along with blood tests to measure erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP) levels at baseline and again at week 24. However, two patients did not participate in the second assessment because they had withdrawn from the study.

Clinical Assessments

The clinical assessments conducted included the 28-joint Swollen Joint Count (SJC, scored from 0 to 28), 28-joint Tender Joint Count (TJC, scored from 0 to 28), general health (GH, scored from 0 [very well] to 100 [worst possible]), Patient Global Assessment (PGA, scored from 0 [very well] to 10 [worst possible]), Physician Global Assessment (PhGA, scored from 0 [very well] to 10 [worst possible]), 28-joint Disease Activity Score based on ESR (DAS28-ESR), 28-joint Disease Activity Score based on CRP (DAS28-CRP), Clinical Disease Activity Index (CDAI), and Simplified Disease Activity Index (SDAI). The 28 joints assessed included the bilateral shoulders, elbows, wrists, metacarpophalangeal (MCP) joints 1-5, thumb interphalangeal joints, proximal interphalangeal joints 2-5, and knees. An experienced rheumatologist performed these clinical assessments. The DAS28-ESR, DAS28-CRP, CDAI, and SDAI were calculated using universally recognized formulas [13,14]:

(1) DAS28-ESR=0.56×√TJC+0.28×√SJC+0.70×ln(ESR)+0.014×GH
(2) DAS28-CRP=0.56×√TJC+0.28×√SJC+0.36×ln(CRP+1)+0.014× GH+0.96
(3) CDAI=SJC+TJC+PGA+PhGA
(4) SDAI=SJC+TJC+PGA+PhGA+CRP

The patients exhibited various levels of RA disease activity, from clinical remission to moderate activity, with DAS28-ESR scores varying from 0.49 to 4.34 (Table 1). The RA patients were categorized into three subgroups: deep clinical remission (dCR, defined as DAS28-ESR ≤1.98), mild clinical remission (mCR, defined as DAS28-ESR >1.98 and <2.6), and low-to-moderate disease activity (LMDA, defined as DAS28-ESR ≥2.6 and ≤5.1) [15].

Baseline characteristics of the 33 included rheumatoid arthritis patients

Ultrasound Assessments

Ultrasound examinations of both the right and left wrist joints were conducted using two different ultrasound imaging systems: the Prodigy programmable platform (S-Sharp Corporation, New Taipei City, Taiwan) and the Philips iU22 machine (Philips Ultrasound, Bothell, WA, USA). Each system was equipped with linear transducers: the Prodigy platform utilized a 7.5-MHz probe, while the Philips machine used a 5- to 12-MHz probe. To prevent movement during measurements, which could cause clutter, these transducers were secured with a bracket. Ultrafast data acquisition on the Prodigy ultrasound system involved the following parameters: a center frequency of 7.5 MHz, a plane-wave emission rate of 1,024 waves/s across eight steering angles (-7°, -5°, -3°, -1°, +1°, +3°, +5°, and +7°), 64 frames, three cycles, and a recording duration of 1.5 seconds. The sequential emission of eight plane waves at different steering angles resulted in the production of one compounded frame. A total of 64 compound frames were produced per cycle at a frame rate of 128 Hz. An ensemble size of 16 was used to generate the PD pixels, with a 50% overlap between consecutive ensembles (Fig. 1). To eliminate both tissue and noise signals in the generation of ultrafast PD images, an SVD filter based on spatial singular vectors was used [16]. The following settings were used for conventional PD with the Philips iU22 machine: center frequency of 6.5 MHz, PRF of 500 Hz, and a wall filter with a cutoff of 47 Hz. Image quality was optimized by finely adjusting the color gain until colored pixels were no longer visible beneath the bone cortex.

Fig. 1.

Design of ultrafast power Doppler data acquisition.

In each cycle, 512 plane waves were emitted at a frequency of 1,024 waves per second. Eight plane waves, with angles varying from -7° to 7°, were subjected to a compounding beamforming process to generate a compound frame. Over three cycles, a total of 64 compound frames were produced within 1.5 s. A video was created by grouping these compound frames into subgroups (for example, every 16 compound frames formed one Doppler frame in the slow-time window), with an overlap ratio of 50% in the window. The resulting Doppler frame rate was 16 frames per second.

The room temperature was maintained at 25°C during the ultrasound assessments. Participants were instructed to rest, which included a 15-minute period of joint immobility to prevent exercise-induced articular hyperemia. The wrist was positioned on the table in a standardized manner and scanned longitudinally across the radiolunate and lunocapitate joints [17]. Accurate vascularity assessments required that the transducer did not excessively compress the skin, in order to minimize the risk of generating false negatives, with ideally a thin gel layer applied between the transducer and skin surface. The author (Lai KL), an experienced rheumatologist with 18 years of experience in ultrasonography, performed the ultrasound assessments.

Principles of SVD Filtering

The ultrasound data, which were initially represented in a 3D space as s(x, z, t), were transformed into a two-dimensional space-time Casorati matrix, denoted as S [18]. This matrix, S, was then decomposed into spatial and temporal singular vectors, accompanied by a diagonal matrix. Estimators used these spatial singular vectors to represent S as a combination of separable image intensity matrices, each scaled by corresponding diagonal coefficients. Tissue signals, characterized by higher spatial-temporal coherence, predominantly occupied the early singular vectors, while blood signals were more aligned with the later vectors. The SVD filter selectively removed these early singular vectors, effectively filtering out the tissue signals but preserving the blood signals [6,8].

The perfusion subspace is characterized by very slow capillary flows, which overlap with the tissue subspace. This subspace can be distinguished based on the physiological characteristics of the slow-flow blood and tissue composition. Noninflamed soft tissue can typically be approximated as consisting of 36% dry tissue, 8.3% slow-flow blood, and 55.7% static water [19,20]. Since static water does not produce echoes, the returned echoes are primarily from dry tissue and slow-flow blood. Notably, the signal intensity from slow-flow blood accounts for 18.7% of the total signal intensity (Fig. 2). In the case of inflamed soft tissue, the perfusion image is generated from signals of both slow-flow blood, which contribute 18.7% to the intensity of the tissue-perfusion subspace, and fast-flow blood signals (Fig. 3).

Fig. 2.

Distributions of tissue and perfusion signals from a wrist with rheumatoid arthritis exhibiting inactive disease.

A. Capillary flow signals (P) are shown in the spatial singular value decomposition (SVD) correlation matrix. No fast-flow signal is evident. B. Ultrafast power Doppler (PD) images: (B1) SVD orders 1-33 represent tissue signals (T), and (B2) SVD orders 34-49 represent P. C. The final PD image was generated by overlaying the compound grayscale image with ultrafast PD image B2.

Fig. 3.

Distributions of tissue, perfusion, and fast-flow signals from a wrist with rheumatoid arthritis exhibiting active disease.

A. Fast-flow signals (F) are present in addition to capillary flow signals P in the spatial singular value decomposition (SVD) correlation matrix. B. Ultrafast power Doppler (PD) images: (B1) SVD orders 1-28 represent tissue signals T, (B2) SVD orders 29-43 represent P, (B3) SVD orders 44-87 represent F, and (B4) SVD orders 29-87 represent P and F. C. The final PD image was generated by overlaying the compound grayscale image with ultrafast PD image B4.

Additionally, noise components that aligned with the final singular vectors were removed using a double-threshold approach [16]. Subsequently, a high-quality PD image was generated from the entire dataset by integrating the energy of the high-frequency signal (x, z, t) in each pixel [8,21].

Image Analysis

The semiquantitative assessments of synovitis were based on the criteria established by OMERACT. The results were represented by a GS score, which indicates the degree of synovial hypertrophy on a scale from 0 to 3, and a PD score, also ranging from 0 to 3 [9].

Quantitative assessments of the blood flow area within the synovium, referred to as the synovial PD area, were conducted using both conventional and ultrafast PD images. An ultrafast video, consisting of 23 PD frames captured at a rate of 16 frames per second, was analyzed. A multiframe image analysis approach was employed to assess the dynamic variations in blood volume and the maximum degree of hyperemia within the synovium. Our analysis focused on the odd-numbered frames (frames 1, 3, 5, ..., 23; totaling 12 frames). The region of interest within the synovium was manually delineated on a B-mode compound image by an experienced rheumatologist (Lai KL) and used to segment the colored pixels in the PD frames. The segmented PD frames were then converted into GS images, which were used for quantitative assessments of the synovial PD brightness (ranging from 0 to 255 for each pixel, with the sum of GS values calculated across all pixels), synovial PD area (referring to the blood flow area, with a binary threshold of 0.05), synovial dilated vessels (referring to arterioles and venules, with a binary threshold of 0.75), and synovial capillaries (identified by subtracting images of the synovial dilated vessels from those of the synovial PD area) (Fig. 4). The average synovial PD brightness was determined by calculating the brightness across the 12 PD frames, while the peak synovial PD brightness was identified as the maximum brightness observed among these frames. All tasks related to the quantitative image analysis were performed using MATLAB software (version R2018b, The MathWorks, Natick, MA, USA).

Fig. 4.

Algorithm of the multiframe image analysis of the ultrafast power Doppler (PD) video.

Each video had 23 frames, with the odd frames being used in the analysis. The region of interest (ROI) of the synovium was delineated manually on a B-mode compound image and used for segmenting the colored pixels in the PD frames. The segmented PD frames were converted to grayscale images for subsequent quantitative assessments of the synovial PD brightness, synovial PD area, synovial dilated vessels, and synovial capillaries.

Statistical Analyses

Continuous variables are presented as mean±standard deviation (range) or median (range) values and were compared using the Kruskal-Wallis test. Categorical variables were analyzed using the chi-square test, with Fisher’s exact test applied when appropriate. All statistical analyses were performed in a two-tailed manner, considering P-values <0.05 as statistically significant. The correlations between PD variables and clinical indices were assessed using the Pearson correlation coefficient (r). These correlations were categorized as none (r<0.1), weak (0.1≤r<0.3), moderate (0.3≤r<0.6), or strong (r≥0.6). Statistical analyses were conducted using SPSS version 18.0 (SPSS Inc., Chicago, IL, USA).

Results

The number of colored pixels in the synovial region was larger in the ultrafast PD images compared to the conventional PD images (Fig. 5). Additionally, the ultrafast PD video exhibited a flicker of colored pixels. Data comparing the ultrafast and conventional PD variables for wrists across different RA disease activity statuses are presented in Table 2. In conventional PD imaging, there were no significant differences in either the OMERACT PD score or the synovial PD area among the RA subgroups. Conventional PD imaging resulted in a large proportion (≥50.0%) of cases with a PD score of 0, including those in the LMDA subgroup. Conversely, ultrafast PD imaging produced only one case with a PD score of 0. In the ultrafast PD imaging, the OMERACT PD score, synovial PD area, area of synovial capillaries, and average synovial PD brightness were all significantly higher in the LMDA group compared to the dCR subgroup, with P-values of 0.005, 0.04, 0.03, and 0.02, respectively.

Fig. 5.

A 48-year-old woman with RA exhibiting mild disease activity.

A. Conventional power Doppler (PD) shows a colored dot (121 pixels) at the periphery of the synovium (*). B. Ultrafast PD shows extensive coloring (1,696 pixels) in the confluence at both the periphery and base of the synovium (* and #). R, radius; L, lunate; C, capitate; T, tendon.

Ultrafast and conventional PD variables of the wrists for different rheumatoid arthritis disease activity statuses

The correlations of all quantitative ultrafast PD variables with DAS28-CRP, CDAI, and SDAI were moderately positive (all P<0.001, with the highest r-value [0.374] observed between the synovial average PD brightness and DAS28-CRP/SDAI). The correlations between quantitative ultrafast PD variables (except for synovial dilated vessels) and DAS28-ESR were weakly positive (all P<0.05). Additionally, the correlation between synovial peak PD brightness and CRP was weakly positive (P<0.05). The ultrafast OMERACT PD score showed weak correlations with all clinical indices except for CRP. Conventional PD variables demonstrated moderately positive correlations with DAS28-CRP, CDAI, and SDAI (all P<0.001, with the highest r-value [0.340] observed between the OMERACT PD score and DAS28-CRP), but they did not correlate with other clinical indices. Synovial hypertrophy, as measured by the OMERACT GS score, showed a moderate correlation with DAS28-CRP (r=0.330, P<0.001) and weak correlations with CRP, DAS28-ESR, CDAI, and SDAI (r=0.230, r=0.210, r=0.228, and r=0.250, respectively; all P<0.05) (Table 3).

Correlations between PD variables and clinical disease activity indices for the entire rheumatoid arthritis population

For patients with dCR, the coefficients showed strong positive correlations between synovial peak PD brightness and DAS28-CRP, CDAI, and SDAI (all P<0.001, with the highest r-value [0.641] between synovial peak PD brightness and DAS28-CRP). Similarly, strong positive correlations were observed between synovial average PD brightness and both CDAI and SDAI (both P<0.001). Moderate positive correlations were noted for synovial peak PD brightness with CRP, synovial average PD brightness with DAS28-CRP, and other ultrafast PD variables with DAS28-CRP, CDAI, and SDAI. Conventional PD variables also demonstrated strong positive correlations with CRP, DAS28-CRP, CDAI, and SDAI (all P<0.001, with the highest r-value [0.867] between synovial PD area and DAS28-CRP). Additionally, there was a moderately positive correlation between the conventional OMERACT PD score and DAS28-ESR (P<0.05). The OMERACT GS score showed moderate correlations with both CRP and DAS28-CRP (both P<0.01) (Table 4).

Correlations between PD variables and clinical disease activity indices for rheumatoid arthritis patients in dCR

Discussion

Joints often exhibit subclinical synovitis in the early stages of the condition or during the clinical remission of RA. In these phases, the capillaries within the synovium proliferate, whereas larger vessels such as arterioles and venules are either absent or sparse. The volume of capillary blood in the synovium escalates with the intensity of inflammation [22]. Additionally, the synovium may not appear swollen even after the onset of synovial hyperemia [23]. Due to the very low capillary flow rate, visualizing this requires a highly sensitive, ultrafast Doppler system [8]. In more advanced stages of synovitis or during active RA, larger vessels begin to form in the synovium, exhibiting a higher flow rate. An increased level of inflammation leads to a greater number of these newly formed larger vessels [22]. The lower sensitivity of conventional PD limits its ability to detect blood flow only in these later stages. Furthermore, the extent of synovial swelling can vary from mild to severe and correlates directly with the level of inflammation.

The parameters of the ultrafast Doppler used in this study achieved a minimum detectable flow rate of 0.22 mm/s in our phantom experiment, although these data are not shown in the Results section. This indicates that the ultrafast Doppler system was capable of detecting slow flows. We chose the singular order of the perfusion subspace, which accounted for 18.7% of the image intensity in the tissue-perfusion space. This approach is grounded in fundamental physiological principles. The resulting perfusion video displayed flickering colored pixels, reflecting the variability of the perfusion blood flows. In contrast, many previous studies have focused solely on the singular order of the fast-flow subspace to generate ultrafast Doppler images [6,16,21].

This study aimed to evaluate the overall clinical disease activity of RA by imaging a single wrist joint. RA is a polyarticular disease characterized by symmetrical bilateral involvement, affecting most joints to varying degrees. Ideally, an assessment tool would estimate the extent of whole-body articular inflammation from observations of a single joint. Given that RA commonly targets the wrist among all limb joints, we chose the wrist for our experiments. In the entire RA population, both ultrafast and conventional PD variables demonstrated unsatisfactory correlations with clinical indices. This outcome may be attributed to our reliance on imaging data from a single joint. Similarly, a previous study that utilized conventional PD on a single joint (MCP2, MCP3, MCP4, or MCP5) also reported low correlation coefficients ranging from 0.274 to 0.399 between the OMERACT PD score and clinical indices [10]. In general, the summed ultrasound scores for multiple joints are more representative of overall disease activity [24].

This study used Philips conventional PD as the current gold standard for clinical imaging. The statistical analysis showed that both ultrafast PD and Philips conventional PD exhibited similar correlation coefficients with clinical disease activity. This suggests that ultrafast PD is equally effective as Philips conventional PD in evaluating clinical disease activity. Due to its high sensitivity in detecting slow blood flows, ultrafast PD can accurately assess synovial perfusion status. This capability allows ultrafast PD to measure the extent of synovial inflammation even in patients with very low disease activity. In contrast, the use of conventional PD often resulted in a large number of cases with a PD score of 0, particularly in the LMDA subgroup, which limits the clinical utility of this method. Furthermore, despite higher correlation coefficients between conventional PD variables and CRP, DAS28-CRP, CDAI, and SDAI in our dCR subgroup, 61.4% of cases still recorded a PD score of 0. This underscores the limitations in the diagnostic value of conventional PD.

The present study found that the correlations between ultrafast PD variables and composite clinical indices ranged from weak to moderate across the entire RA population, while they were moderate to high in the dCR subgroup. In ultrafast perfusion imaging, the colored pixels predominantly originated from perfusion blood signals rather than fast blood signals, due to the typically larger volume of perfusion blood compared to fast blood. Given this characteristic, ultrafast perfusion imaging proves sensitive in dCR joints, where perfusion blood constitutes the majority of the blood volume. However, it is less sensitive to increases in fast blood volume in joints exhibiting significant disease activity. Furthermore, ultrafast PD variables showed stronger correlations with DAS28-CRP, CDAI, and SDAI than with ESR, CRP, and DAS28-ESR. In clinical practice, DAS28-CRP, CDAI, and SDAI are generally regarded as more accurate indicators of disease activity and have greater relevance to prognoses [14,25]. These observations suggest that ultrafast PD can be effectively utilized in clinical settings to assess disease activity and therapeutic efficacy, particularly in RA patients in dCR.

Nonetheless, this study had some limitations. First, a major challenge in SVD thresholding is the overlap between perfusion and tissue signals. Relying solely on physiological data for thresholding may not be the most effective approach. One potential improvement could be the adaptive demodulation of tissue signals [26]. Second, the relationship between the value of an image pixel, signal energy, and blood volume is not linear. Consequently, the binary threshold of 0.75 used for segmenting arterioles and venules was determined empirically and may not be precise. Further research into methods to enhance segmentation accuracy is necessary. Third, the correlations between ultrafast PD variables and clinical indices were only weak to moderate across the entire RA population. Physicians should therefore be cautious when using ultrafast PD image data to diagnose and treat RA patients. Fourth, focusing solely on the wrist in this study may be a limitation. It is generally understood that evaluating the most severely or acutely affected joint, rather than just the wrist, could yield a stronger correlation with clinical disease activity in individual patients.

The proliferation of the capillary network marks the initial stage of angiogenesis in RA. This study has shown that ultrafast PD imaging can extract signals of capillary flows by selecting appropriate singular vectors and generating perfusion images. Ultrafast PD variables derived from single-wrist perfusion images exhibited weak-to-moderate correlations with clinical indices across the entire RA population, with stronger correlations observed in the dCR subgroup. These findings suggest that ultrafast PD perfusion imaging could be useful for assessing RA disease activity, particularly in patients in dCR. Future studies might focus on clinical correlations in the early stages of RA, as this approach could be more practical and relevant for clinical applications than focusing on RA in dCR.

Notes

Author Contributions

Conceptualization: Lai KL, Li PC. Data acquisition: Lai KL. Data analysis or interpretation: Lai KL, Li PC. Drafting of the manuscript: Lai KL. Critical revision of the manuscript: Li PC. Approval of the final version of the manuscript: all authors.

No potential conflict of interest relevant to this article was reported.

Acknowledgements

This work was supported by Taichung Veterans General Hospital Research Program (protocol number: TCVGH-1123804D) and the "Center for Advanced Computing and Imaging in Biomedicine (grant numbers: NTU-113L900703)" from The Featured Areas Research Center Program within the framework of the Higher Education Sprout Project by the Ministry of Education (MOE) in Taiwan.

References

1. Rooney M, Condell D, Quinlan W, Daly L, Whelan A, Feighery C, et al. Analysis of the histologic variation of synovitis in rheumatoid arthritis. Arthritis Rheum 1988;31:956–963.
2. Colebatch AN, Edwards CJ, Ostergaard M, van der Heijde D, Balint PV, D'Agostino MA, et al. EULAR recommendations for the use of imaging of the joints in the clinical management of rheumatoid arthritis. Ann Rheum Dis 2013;72:804–814.
3. Lai KL, Chen DY, Wen MC, Chen YM, Hung WT, Chen YH, et al. What does power Doppler signal indicate in rheumatoid synovitis? A point of view from synovial histopathology. J Chin Med Assoc 2018;81:383–386.
4. Meyer MF, Czaplewski H, Braun J, Hellmich B, Schatz H, Klein HH. Lack of a difference in increased capillary blood cell velocity in the skin over proximal interphalangeal joints between rheumatoid arthritis and osteoarthritis. Microvasc Res 2005;70:1–6.
5. Bercoff J, Montaldo G, Loupas T, Savery D, Meziere F, Fink M, et al. Ultrafast compound Doppler imaging: providing full blood flow characterization. IEEE Trans Ultrason Ferroelectr Freq Control 2011;58:134–147.
6. Demene C, Deffieux T, Pernot M, Osmanski BF, Biran V, Gennisson JL, et al. Spatiotemporal clutter filtering of Ultrafast ultrasound data highly increases Doppler and fUltrasound Sensitivity. IEEE Trans Med Imaging 2015;34:2271–2285.
7. van der Ven M, Luime JJ, van der Velden LL, Bosch JG, Hazes JM, Vos HJ. High-frame-rate power Doppler ultrasound is more sensitive than conventional Power doppler in detecting rheumatic vascularisation. Ultrasound Med Biol 2017;43:1868–1879.
8. Lai KL, Tsai MC, Li PC. Ultrafast Doppler imaging for early detection of synovitis in rheumatoid arthritis. Ultrasound Med Biol 2024;50:484–493.
9. D'Agostino MA, Terslev L, Aegerter P, Backhaus M, Balint P, Bruyn GA, et al. Scoring ultrasound synovitis in rheumatoid arthritis: a EULAR-OMERACT ultrasound taskforce-Part 1: definition and development of a standardised, consensus-based scoring system. RMD Open 2017;3:e000428.
10. Deng X, Sun X, Xie W, Wang Y, Zhang Z. The correlation of ultrasound-detected synovitis in an individual small joint with overall clinical disease activity in patients with rheumatoid arthritis. Clin Rheumatol 2022;41:3319–3324.
11. Yen TH, Wu YD, Chen HH, Tseng CW, Chen YH, Chen YM, et al. The role of ultrasound synovitis scores for patients who are at risk of rheumatoid arthritis. Int J Rheum Dis 2023;26:922–929.
12. Aletaha D, Neogi T, Silman AJ, Funovits J, Felson DT, Bingham CO 3rd, et al. 2010 rheumatoid arthritis classification criteria: an American College of Rheumatology/European League Against Rheumatism collaborative initiative. Ann Rheum Dis 2010;69:1580–1588.
13. Aletaha D, Ward MM, Machold KP, Nell VP, Stamm T, Smolen JS. Remission and active disease in rheumatoid arthritis: defining criteria for disease activity states. Arthritis Rheum 2005;52:2625–2636.
14. Anderson JK, Zimmerman L, Caplan L, Michaud K. Measures of rheumatoid arthritis disease activity: Patient (PtGA) and Provider (PrGA) Global Assessment of Disease Activity, Disease Activity Score (DAS) and Disease Activity Score with 28-Joint Counts (DAS28), Simplified Disease Activity Index (SDAI), Clinical Disease Activity Index (CDAI), Patient Activity Score (PAS) and Patient Activity Score-II (PASII), Routine Assessment of Patient Index Data (RAPID), Rheumatoid Arthritis Disease Activity Index (RADAI) and Rheumatoid Arthritis Disease Activity Index-5 (RADAI-5), Chronic Arthritis Systemic Index (CASI), Patient-Based Disease Activity Score With ESR (PDAS1) and Patient-Based Disease Activity Score without ESR (PDAS2), and Mean Overall Index for Rheumatoid Arthritis (MOI-RA). Arthritis Care Res (Hoboken) 2011;63 Suppl 11:S14–S36.
15. Geng Y, Han J, Deng X, Zhang Z. Deep clinical remission: an optimised target in the management of rheumatoid arthritis? Experience from an ultrasonography study. Clin Exp Rheumatol 2016;34:581–586.
16. Baranger J, Arnal B, Perren F, Baud O, Tanter M, Demene C. Adaptive spatiotemporal SVD clutter filtering for Ultrafast Doppler imaging using similarity of spatial singular vectors. IEEE Trans Med Imaging 2018;37:1574–1586.
17. Backhaus M, Burmester GR, Gerber T, Grassi W, Machold KP, Swen WA, et al. Guidelines for musculoskeletal ultrasound in rheumatology. Ann Rheum Dis 2001;60:641–649.
18. Candes EJ, Sing-Long CA, Trzasko JD. Unbiased risk estimates for singular value thresholding and spectral estimators. IEEE Trans Signal Process 2013;61:4643–4657.
19. Mitchell HH, Hamilton TS, Steggerda FR, Bean HW. The chemical composition of the adult human body and its bearing on the biochemistry of growth. J Biol Chem 1945;158:625–637.
20. Keith NM, Rowntree LG, Geraghty JT. A method for the determination of plasma and blood volume. Arch Intern Med 1915;16:547–576.
21. Demene C, Pernot M, Biran V, Alison M, Fink M, Baud O, et al. Ultrafast Doppler reveals the mapping of cerebral vascular resistivity in neonates. J Cereb Blood Flow Metab 2014;34:1009–1017.
22. Balogh E, Biniecka M, Fearon U, Veale DJ, Szekanecz Z. Angiogenesis in inflammatory arthritis. Isr Med Assoc J 2019;21:345–352.
23. Horie T, Nishida M, Tanimura S, Kamishima T, Tamai E, Morimura Y, et al. Detection of increased vascular signal in arthritis-prone rats without joint swelling using superb microvascular imaging ultrasonography. Ultrasound Med Biol 2019;45:2086–2093.
24. Zhu J, Wang J, Dong Y, Song Y, Huang P. Correlation between ultrasonographic scores and American college of rheumatology recommended rheumatoid arthritis disease activity measures: a systematic review and network meta-analysis. J Ultrasound 2023;26:39–47.
25. Balsa A, de Miguel E, Castillo C, Peiteado D, Martin-Mola E. Superiority of SDAI over DAS-28 in assessment of remission in rheumatoid arthritis patients using power Doppler ultrasonography as a gold standard. Rheumatology (Oxford) 2010;49:683–690.
26. Tierney J, Walsh K, Griffith H, Baker J, Brown DB, Byram B. Combining slow flow techniques with adaptive demodulation for improved perfusion ultrasound imaging without contrast. IEEE Trans Ultrason Ferroelectr Freq Control 2019;66:834–848.

Article information Continued

Notes

Key point

Ultrafast power Doppler (PD) imaging with singular value decomposition filtering can be used to detect synovial perfusion, which represents the inflammatory burden of rheumatoid arthritis (RA). The variables measured in ultrafast PD perfusion imaging exhibit various correlations with clinical indices of RA disease activity, with stronger correlations among patients in deep clinical remission.

Fig. 1.

Design of ultrafast power Doppler data acquisition.

In each cycle, 512 plane waves were emitted at a frequency of 1,024 waves per second. Eight plane waves, with angles varying from -7° to 7°, were subjected to a compounding beamforming process to generate a compound frame. Over three cycles, a total of 64 compound frames were produced within 1.5 s. A video was created by grouping these compound frames into subgroups (for example, every 16 compound frames formed one Doppler frame in the slow-time window), with an overlap ratio of 50% in the window. The resulting Doppler frame rate was 16 frames per second.

Fig. 2.

Distributions of tissue and perfusion signals from a wrist with rheumatoid arthritis exhibiting inactive disease.

A. Capillary flow signals (P) are shown in the spatial singular value decomposition (SVD) correlation matrix. No fast-flow signal is evident. B. Ultrafast power Doppler (PD) images: (B1) SVD orders 1-33 represent tissue signals (T), and (B2) SVD orders 34-49 represent P. C. The final PD image was generated by overlaying the compound grayscale image with ultrafast PD image B2.

Fig. 3.

Distributions of tissue, perfusion, and fast-flow signals from a wrist with rheumatoid arthritis exhibiting active disease.

A. Fast-flow signals (F) are present in addition to capillary flow signals P in the spatial singular value decomposition (SVD) correlation matrix. B. Ultrafast power Doppler (PD) images: (B1) SVD orders 1-28 represent tissue signals T, (B2) SVD orders 29-43 represent P, (B3) SVD orders 44-87 represent F, and (B4) SVD orders 29-87 represent P and F. C. The final PD image was generated by overlaying the compound grayscale image with ultrafast PD image B4.

Fig. 4.

Algorithm of the multiframe image analysis of the ultrafast power Doppler (PD) video.

Each video had 23 frames, with the odd frames being used in the analysis. The region of interest (ROI) of the synovium was delineated manually on a B-mode compound image and used for segmenting the colored pixels in the PD frames. The segmented PD frames were converted to grayscale images for subsequent quantitative assessments of the synovial PD brightness, synovial PD area, synovial dilated vessels, and synovial capillaries.

Fig. 5.

A 48-year-old woman with RA exhibiting mild disease activity.

A. Conventional power Doppler (PD) shows a colored dot (121 pixels) at the periphery of the synovium (*). B. Ultrafast PD shows extensive coloring (1,696 pixels) in the confluence at both the periphery and base of the synovium (* and #). R, radius; L, lunate; C, capitate; T, tendon.

Table 1.

Baseline characteristics of the 33 included rheumatoid arthritis patients

Characteristic Value
Female sex 28 (84.8)
Age (year) 49.2±12.4 (21-71)
Seropositivity 25 (75.8)
ESR (mm/h) 14.0±18.7 (1-100)
CRP (mg/dL) 0.470±1.023 (0.01-5.276)
SJC 0.8±1.2 (0-5)
TJC 0.6±0.8 (0-2)
PGA 3.0±2.2 (0-7)
PhGA 1.7±1.5 (0-4)
DAS28-ESR 2.31±1.03 (0.49-4.34)
DAS28-CRP 2.09±0.76 (1.00-4.42)
CDAI 5.2±3.9 (0-14)
SDAI 6.531±4.883 (0.024-19.276)

Value are presented as number (%) or mean±standard deviation (range).

ESR, erythrocyte sedimentation rate; CRP, C-reactive protein; SJC, Swollen Joint Count (scored from 0 to 28); TJC, Tender Joint Count (scored from 0 to 28); PGA, Patient Global Assessment (scored from 0 to 10); PhGA, Physician Global Assessment (scored from 0 to 10); DAS28-ESR, 28-joint Disease Activity Score based on ESR; DAS28-CRP, 28-joint Disease Activity Score based on CRP; CDAI, Clinical Disease Activity Index; SDAI, Simplified Disease Activity Index.

Table 2.

Ultrafast and conventional PD variables of the wrists for different rheumatoid arthritis disease activity statuses

dCR (n=44) mCR (n=38) LMDA (n=46) P-value (dCR vs. mCR) P-value (mCR vs. LMDA) P-value (dCR vs. LMDA)
Conventional PD
 OMERACT PD score 0 (0-3) 0 (0-2) 0.5 (0-3) 0.716 0.324 0.176
 No. of joints with a PD score of 0 27 (61.4) 21 (55.3) 23 (50.0) 0.576 0.631 0.278
 Synovial PD area, pixels 0 (0-4,513.6) 17.3 (0-3,977) 22.1 (0-7,686.6) 0.949 0.151 0.169
Ultrafast PD
 OMERACT PD score 2 (0-3) 2 (1-3) 2 (1-3) 0.339 0.037 0.005
 No. of joints with a PD score of 0 1 (2.3) 0 0 0.537 >0.999 0.489
 Synovial PD area, pixels 13,323.5 (2,477.1-35,515.8) 16,143.1 (3,824.9-27,569.3) 16,258.3 (2,051.2-39,338.8) 0.131 0.430 0.036
 Synovial dilated vessels, pixels 135.4 (0-1,335.3) 152.6 (0-1,313.5) 201.1 (0-1,391) 0.594 0.149 0.408
 Synovial capillaries, pixels 13,077.5 (2,472.9-35,127.7) 15,963.4 (3,723.5-27,095.6) 15,661.8 (2,051.2-39,256.9) 0.116 0.453 0.034
 Average synovial PD brightness 861,360.1 (249,360.7-2,311,422) 1,019,475.8 (256,855.3-2,217,110.9) 1,115,316.5 (209,440.7-2,474,321) 0.146 0.320 0.018
 Peak synovial PD brightness 1,426,835 (470,016-3,535,580) 1,760,747.5 (583,621-3,617,455) 1,774,772.5 (384,495-4,247,904) 0.127 0.643 0.057

Values are presented as median (range) or number (%).

Note: The number represents the number of joints observed. This trial included 33 patients, each of whom underwent an ultrasound examination of both wrist joints at weeks 0 and 24. However, two patients did not return for the week 24 examination, resulting in a total of 128 joint examinations performed.

Ultrafast parameters: center frequency=7.5 MHz, pulse-repetition frequency=1,024 Hz, 8 angles (-7°, -5°, -3°, -1°, 1°, 3°, 5°, and 7°), 64 frames, 3 cycles, recording time=1.5 s, ensemble size=16, and overlap=50%.

dCR was defined as DAS28-ESR ≤1.98, mCR was defined as DAS28-ESR >1.98 and <2.6, and LMDA was defined as DAS28-ESR ≥2.6 and ≤5.1.

PD, power Doppler; dCR, deep clinical remission; mCR, mild clinical remission; LMDA, low-to-moderate disease activity; OMERACT, Outcome Measures in Rheumatology; DAS28-ESR, 28-joint Disease Activity Score based on erythrocyte sedimentation rate.

Table 3.

Correlations between PD variables and clinical disease activity indices for the entire rheumatoid arthritis population

r-value (P-value)
ESR CRP DAS28-ESR DAS28-CRP CDAI SDAI
Ultrafast PD
 Peak synovial PD brightness 0.139 (0.119) 0.224 (0.011) 0.232 (0.008) 0.364 (<0.001) 0.343 (<0.001) 0.353 (<0.001)
 Average synovial PD brightness 0.124 (0.163) 0.192 (0.030) 0.228 (0.010) 0.374 (<0.001) 0.373 (<0.001) 0.374 (<0.001)
 Synovial PD area 0.171 (0.054) 0.152 (0.087) 0.221 (0.012) 0.321 (<0.001) 0.328 (<0.001) 0.326 (<0.001)
 Synovial dilated vessels -0.114 (NA) 0.150 (0.090) 0.064 (0.475) 0.313 (<0.001) 0.323 (<0.001) 0.322 (<0.001)
 Synovial capillaries 0.178 (0.044) 0.149 (0.093) 0.223 (0.012) 0.315 (<0.001) 0.322 (<0.001) 0.320 (<0.001)
 OMERACT PD score 0.235 (0.008) 0.086 (0.337) 0.254 (0.004) 0.294 (<0.001) 0.298 (<0.001) 0.287 (0.001)
Conventional PD
 Synovial PD area -0.093 (NA) 0.105 (0.240) 0.155 (0.080) 0.363 (<0.001) 0.337 (<0.001) 0.326 (<0.001)
 OMERACT PD score -0.076 (NA) 0.107 (0.228) 0.161 (0.069) 0.340 (<0.001) 0.312 (<0.001) 0.304 (<0.001)
 OMERACT GS score 0.000 (0.999) 0.230 (0.009) 0.210 (0.017) 0.330 (<0.001) 0.228 (0.009) 0.250 (0.004)

PD, power Doppler; ESR, erythrocyte sedimentation rate; CRP, C-reactive protein; DAS28-ESR, 28-joint Disease Activity Score based on ESR; DAS28-CRP, 28-joint Disease Activity Score based on CRP; CDAI, Clinical Disease Activity Index; SDAI, Simplified Disease Activity Index; NA, not applicable; OMERACT, Outcome Measures in Rheumatology; GS, grayscale.

Table 4.

Correlations between PD variables and clinical disease activity indices for rheumatoid arthritis patients in dCR

r-value (P-value)
ESR CRP DAS28-ESR DAS28-CRP CDAI SDAI
Ultrafast PD
 Peak synovial PD brightness -0.476 (NA) 0.381 (0.011) -0.022 (NA) 0.641 (<0.001) 0.627 (<0.001) 0.630 (<0.001)
 Average synovial PD brightness -0.524 (NA) 0.261 (0.087) -0.114 (NA) 0.578 (<0.001) 0.615 (<0.001) 0.611 (<0.001)
 Synovial PD area -0.493 (NA) 0.227 (0.139) -0.111 (NA) 0.546 (<0.001) 0.579 (<0.001) 0.576 (<0.001)
 Synovial dilated vessels -0.425 (NA) 0.248 (0.105) -0.075 (NA) 0.491 (<0.001) 0.505 (<0.001) 0.504 (<0.001)
 Synovial capillaries -0.490 (NA) 0.223 (0.147) -0.111 (NA) 0.541 (<0.001) 0.575 (<0.001) 0.570 (<0.001)
 OMERACT PD score -0.537 (NA) 0.291 (0.054) -0.099 (NA) 0.492 (<0.001) 0.560 (<0.001) 0.560 (<0.001)
Conventional PD
 Synovial PD area -0.330 (NA) 0.631 (<0.001) 0.260 (0.088) 0.867 (<0.001) 0.774 (<0.001) 0.786 (<0.001)
 OMERACT PD score -0.231 (NA) 0.603 (<0.001) 0.306 (0.043) 0.735 (<0.001) 0.614 (<0.001) 0.629 (<0.001)
 OMERACT GS score -0.182 (NA) 0.493 (<0.001) 0.076 (0.621) 0.427 (0.004) 0.222 (0.147) 0.242 (0.113)

PD, power Doppler; dCR, deep clinical remission; ESR, erythrocyte sedimentation rate; CRP, C-reactive protein; DAS28-ESR, 28-joint Disease Activity Score based on ESR; DAS28-CRP, 28-joint Disease Activity Score based on CRP; CDAI, Clinical Disease Activity Index; SDAI, Simplified Disease Activity Index; NA, not applicable; OMERACT, Outcome Measures in Rheumatology; GS, grayscale.